2 .\" Copyright (c) 2013 by Turbo Fredriksson <turbo@bayour.com>. All rights reserved.
3 .\" The contents of this file are subject to the terms of the Common Development
4 .\" and Distribution License (the "License"). You may not use this file except
5 .\" in compliance with the License. You can obtain a copy of the license at
6 .\" usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing.
8 .\" See the License for the specific language governing permissions and
9 .\" limitations under the License. When distributing Covered Code, include this
10 .\" CDDL HEADER in each file and include the License file at
11 .\" usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this
12 .\" CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your
13 .\" own identifying information:
14 .\" Portions Copyright [yyyy] [name of copyright owner]
15 .TH ZFS-MODULE-PARAMETERS 5 "Nov 16, 2013"
17 zfs\-module\-parameters \- ZFS module parameters
21 Description of the different parameters to the ZFS module.
23 .SS "Module parameters"
30 \fBignore_hole_birth\fR (int)
33 When set, the hole_birth optimization will not be used, and all holes will
34 always be sent on zfs send. Useful if you suspect your datasets are affected
35 by a bug in hole_birth.
37 Use \fB1\fR for on and \fB0\fR (default) for off.
43 \fBl2arc_feed_again\fR (int)
46 Turbo L2ARC warm-up. When the L2ARC is cold the fill interval will be set as
49 Use \fB1\fR for yes (default) and \fB0\fR to disable.
55 \fBl2arc_feed_min_ms\fR (ulong)
58 Min feed interval in milliseconds. Requires \fBl2arc_feed_again=1\fR and only
59 applicable in related situations.
61 Default value: \fB200\fR.
67 \fBl2arc_feed_secs\fR (ulong)
70 Seconds between L2ARC writing
72 Default value: \fB1\fR.
78 \fBl2arc_headroom\fR (ulong)
81 How far through the ARC lists to search for L2ARC cacheable content, expressed
82 as a multiplier of \fBl2arc_write_max\fR
84 Default value: \fB2\fR.
90 \fBl2arc_headroom_boost\fR (ulong)
93 Scales \fBl2arc_headroom\fR by this percentage when L2ARC contents are being
94 successfully compressed before writing. A value of 100 disables this feature.
96 Default value: \fB200\fR.
102 \fBl2arc_max_block_size\fR (ulong)
105 The maximum block size which may be written to an L2ARC device, after
106 compression and other factors. This setting is used to prevent a small
107 number of large blocks from pushing a larger number of small blocks out
110 Default value: \fB16,777,216\fR.
116 \fBl2arc_nocompress\fR (int)
119 Skip compressing L2ARC buffers
121 Use \fB1\fR for yes and \fB0\fR for no (default).
127 \fBl2arc_noprefetch\fR (int)
130 Do not write buffers to L2ARC if they were prefetched but not used by
133 Use \fB1\fR for yes (default) and \fB0\fR to disable.
139 \fBl2arc_norw\fR (int)
142 No reads during writes
144 Use \fB1\fR for yes and \fB0\fR for no (default).
150 \fBl2arc_write_boost\fR (ulong)
153 Cold L2ARC devices will have \fBl2arc_write_nax\fR increased by this amount
154 while they remain cold.
156 Default value: \fB8,388,608\fR.
162 \fBl2arc_write_max\fR (ulong)
165 Max write bytes per interval
167 Default value: \fB8,388,608\fR.
173 \fBmetaslab_aliquot\fR (ulong)
176 Metaslab granularity, in bytes. This is roughly similar to what would be
177 referred to as the "stripe size" in traditional RAID arrays. In normal
178 operation, ZFS will try to write this amount of data to a top-level vdev
179 before moving on to the next one.
181 Default value: \fB524,288\fR.
187 \fBmetaslab_bias_enabled\fR (int)
190 Enable metaslab group biasing based on its vdev's over- or under-utilization
191 relative to the pool.
193 Use \fB1\fR for yes (default) and \fB0\fR for no.
199 \fBmetaslab_debug_load\fR (int)
202 Load all metaslabs during pool import.
204 Use \fB1\fR for yes and \fB0\fR for no (default).
210 \fBmetaslab_debug_unload\fR (int)
213 Prevent metaslabs from being unloaded.
215 Use \fB1\fR for yes and \fB0\fR for no (default).
221 \fBmetaslab_fragmentation_factor_enabled\fR (int)
224 Enable use of the fragmentation metric in computing metaslab weights.
226 Use \fB1\fR for yes (default) and \fB0\fR for no.
232 \fBmetaslabs_per_vdev\fR (int)
235 When a vdev is added, it will be divided into approximately (but no more than) this number of metaslabs.
237 Default value: \fB200\fR.
243 \fBmetaslab_preload_enabled\fR (int)
246 Enable metaslab group preloading.
248 Use \fB1\fR for yes (default) and \fB0\fR for no.
254 \fBmetaslab_lba_weighting_enabled\fR (int)
257 Give more weight to metaslabs with lower LBAs, assuming they have
258 greater bandwidth as is typically the case on a modern constant
259 angular velocity disk drive.
261 Use \fB1\fR for yes (default) and \fB0\fR for no.
267 \fBspa_config_path\fR (charp)
272 Default value: \fB/etc/zfs/zpool.cache\fR.
278 \fBspa_asize_inflation\fR (int)
281 Multiplication factor used to estimate actual disk consumption from the
282 size of data being written. The default value is a worst case estimate,
283 but lower values may be valid for a given pool depending on its
284 configuration. Pool administrators who understand the factors involved
285 may wish to specify a more realistic inflation factor, particularly if
286 they operate close to quota or capacity limits.
288 Default value: \fB24\fR.
294 \fBspa_load_verify_data\fR (int)
297 Whether to traverse data blocks during an "extreme rewind" (\fB-X\fR)
298 import. Use 0 to disable and 1 to enable.
300 An extreme rewind import normally performs a full traversal of all
301 blocks in the pool for verification. If this parameter is set to 0,
302 the traversal skips non-metadata blocks. It can be toggled once the
303 import has started to stop or start the traversal of non-metadata blocks.
305 Default value: \fB1\fR.
311 \fBspa_load_verify_metadata\fR (int)
314 Whether to traverse blocks during an "extreme rewind" (\fB-X\fR)
315 pool import. Use 0 to disable and 1 to enable.
317 An extreme rewind import normally performs a full traversal of all
318 blocks in the pool for verification. If this parameter is set to 0,
319 the traversal is not performed. It can be toggled once the import has
320 started to stop or start the traversal.
322 Default value: \fB1\fR.
328 \fBspa_load_verify_maxinflight\fR (int)
331 Maximum concurrent I/Os during the traversal performed during an "extreme
332 rewind" (\fB-X\fR) pool import.
334 Default value: \fB10000\fR.
340 \fBspa_slop_shift\fR (int)
343 Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space
344 in the pool to be consumed. This ensures that we don't run the pool
345 completely out of space, due to unaccounted changes (e.g. to the MOS).
346 It also limits the worst-case time to allocate space. If we have
347 less than this amount of free space, most ZPL operations (e.g. write,
348 create) will return ENOSPC.
350 Default value: \fB5\fR.
356 \fBzfetch_array_rd_sz\fR (ulong)
359 If prefetching is enabled, disable prefetching for reads larger than this size.
361 Default value: \fB1,048,576\fR.
367 \fBzfetch_max_distance\fR (uint)
370 Max bytes to prefetch per stream (default 8MB).
372 Default value: \fB8,388,608\fR.
378 \fBzfetch_max_streams\fR (uint)
381 Max number of streams per zfetch (prefetch streams per file).
383 Default value: \fB8\fR.
389 \fBzfetch_min_sec_reap\fR (uint)
392 Min time before an active prefetch stream can be reclaimed
394 Default value: \fB2\fR.
400 \fBzfs_arc_dnode_limit\fR (ulong)
403 When the number of bytes consumed by dnodes in the ARC exceeds this number of
404 bytes, try to unpin some of it in response to demand for non-metadata. This
405 value acts as a floor to the amount of dnode metadata, and defaults to 0 which
406 indicates that a percent which is based on \fBzfs_arc_dnode_limit_percent\fR of
407 the ARC meta buffers that may be used for dnodes.
409 See also \fBzfs_arc_meta_prune\fR which serves a similar purpose but is used
410 when the amount of metadata in the ARC exceeds \fBzfs_arc_meta_limit\fR rather
411 than in response to overall demand for non-metadata.
414 Default value: \fB0\fR.
420 \fBzfs_arc_dnode_limit_percent\fR (ulong)
423 Percentage that can be consumed by dnodes of ARC meta buffers.
425 See also \fBzfs_arc_dnode_limit\fR which serves a similar purpose but has a
426 higher priority if set to nonzero value.
428 Default value: \fB10\fR.
434 \fBzfs_arc_dnode_reduce_percent\fR (ulong)
437 Percentage of ARC dnodes to try to scan in response to demand for non-metadata
438 when the number of bytes consumed by dnodes exceeds \fBzfs_arc_dnode_limit\fB.
441 Default value: \fB10% of the number of dnodes in the ARC\fR.
447 \fBzfs_arc_average_blocksize\fR (int)
450 The ARC's buffer hash table is sized based on the assumption of an average
451 block size of \fBzfs_arc_average_blocksize\fR (default 8K). This works out
452 to roughly 1MB of hash table per 1GB of physical memory with 8-byte pointers.
453 For configurations with a known larger average block size this value can be
454 increased to reduce the memory footprint.
457 Default value: \fB8192\fR.
463 \fBzfs_arc_evict_batch_limit\fR (int)
466 Number ARC headers to evict per sub-list before proceeding to another sub-list.
467 This batch-style operation prevents entire sub-lists from being evicted at once
468 but comes at a cost of additional unlocking and locking.
470 Default value: \fB10\fR.
476 \fBzfs_arc_grow_retry\fR (int)
479 After a memory pressure event the ARC will wait this many seconds before trying
482 Default value: \fB5\fR.
488 \fBzfs_arc_lotsfree_percent\fR (int)
491 Throttle I/O when free system memory drops below this percentage of total
492 system memory. Setting this value to 0 will disable the throttle.
494 Default value: \fB10\fR.
500 \fBzfs_arc_max\fR (ulong)
503 Max arc size of ARC in bytes. If set to 0 then it will consume 1/2 of system
504 RAM. This value must be at least 67108864 (64 megabytes).
506 This value can be changed dynamically with some caveats. It cannot be set back
507 to 0 while running and reducing it below the current ARC size will not cause
508 the ARC to shrink without memory pressure to induce shrinking.
510 Default value: \fB0\fR.
516 \fBzfs_arc_meta_limit\fR (ulong)
519 The maximum allowed size in bytes that meta data buffers are allowed to
520 consume in the ARC. When this limit is reached meta data buffers will
521 be reclaimed even if the overall arc_c_max has not been reached. This
522 value defaults to 0 which indicates that a percent which is based on
523 \fBzfs_arc_meta_limit_percent\fR of the ARC may be used for meta data.
525 This value my be changed dynamically except that it cannot be set back to 0
526 for a specific percent of the ARC; it must be set to an explicit value.
528 Default value: \fB0\fR.
534 \fBzfs_arc_meta_limit_percent\fR (ulong)
537 Percentage of ARC buffers that can be used for meta data.
539 See also \fBzfs_arc_meta_limit\fR which serves a similar purpose but has a
540 higher priority if set to nonzero value.
543 Default value: \fB75\fR.
549 \fBzfs_arc_meta_min\fR (ulong)
552 The minimum allowed size in bytes that meta data buffers may consume in
553 the ARC. This value defaults to 0 which disables a floor on the amount
554 of the ARC devoted meta data.
556 Default value: \fB0\fR.
562 \fBzfs_arc_meta_prune\fR (int)
565 The number of dentries and inodes to be scanned looking for entries
566 which can be dropped. This may be required when the ARC reaches the
567 \fBzfs_arc_meta_limit\fR because dentries and inodes can pin buffers
568 in the ARC. Increasing this value will cause to dentry and inode caches
569 to be pruned more aggressively. Setting this value to 0 will disable
570 pruning the inode and dentry caches.
572 Default value: \fB10,000\fR.
578 \fBzfs_arc_meta_adjust_restarts\fR (ulong)
581 The number of restart passes to make while scanning the ARC attempting
582 the free buffers in order to stay below the \fBzfs_arc_meta_limit\fR.
583 This value should not need to be tuned but is available to facilitate
584 performance analysis.
586 Default value: \fB4096\fR.
592 \fBzfs_arc_min\fR (ulong)
597 Default value: \fB100\fR.
603 \fBzfs_arc_min_prefetch_lifespan\fR (int)
606 Minimum time prefetched blocks are locked in the ARC, specified in jiffies.
607 A value of 0 will default to 1 second.
609 Default value: \fB0\fR.
615 \fBzfs_arc_num_sublists_per_state\fR (int)
618 To allow more fine-grained locking, each ARC state contains a series
619 of lists for both data and meta data objects. Locking is performed at
620 the level of these "sub-lists". This parameters controls the number of
621 sub-lists per ARC state.
623 Default value: \fR1\fB or the number of online CPUs, whichever is greater
629 \fBzfs_arc_overflow_shift\fR (int)
632 The ARC size is considered to be overflowing if it exceeds the current
633 ARC target size (arc_c) by a threshold determined by this parameter.
634 The threshold is calculated as a fraction of arc_c using the formula
635 "arc_c >> \fBzfs_arc_overflow_shift\fR".
637 The default value of 8 causes the ARC to be considered to be overflowing
638 if it exceeds the target size by 1/256th (0.3%) of the target size.
640 When the ARC is overflowing, new buffer allocations are stalled until
641 the reclaim thread catches up and the overflow condition no longer exists.
643 Default value: \fB8\fR.
650 \fBzfs_arc_p_min_shift\fR (int)
653 arc_c shift to calc min/max arc_p
655 Default value: \fB4\fR.
661 \fBzfs_arc_p_aggressive_disable\fR (int)
664 Disable aggressive arc_p growth
666 Use \fB1\fR for yes (default) and \fB0\fR to disable.
672 \fBzfs_arc_p_dampener_disable\fR (int)
675 Disable arc_p adapt dampener
677 Use \fB1\fR for yes (default) and \fB0\fR to disable.
683 \fBzfs_arc_shrink_shift\fR (int)
686 log2(fraction of arc to reclaim)
688 Default value: \fB5\fR.
694 \fBzfs_arc_sys_free\fR (ulong)
697 The target number of bytes the ARC should leave as free memory on the system.
698 Defaults to the larger of 1/64 of physical memory or 512K. Setting this
699 option to a non-zero value will override the default.
701 Default value: \fB0\fR.
707 \fBzfs_autoimport_disable\fR (int)
710 Disable pool import at module load by ignoring the cache file (typically \fB/etc/zfs/zpool.cache\fR).
712 Use \fB1\fR for yes (default) and \fB0\fR for no.
718 \fBzfs_dbgmsg_enable\fR (int)
721 Internally ZFS keeps a small log to facilitate debugging. By default the log
722 is disabled, to enable it set this option to 1. The contents of the log can
723 be accessed by reading the /proc/spl/kstat/zfs/dbgmsg file. Writing 0 to
724 this proc file clears the log.
726 Default value: \fB0\fR.
732 \fBzfs_dbgmsg_maxsize\fR (int)
735 The maximum size in bytes of the internal ZFS debug log.
737 Default value: \fB4M\fR.
743 \fBzfs_dbuf_state_index\fR (int)
746 This feature is currently unused. It is normally used for controlling what
747 reporting is available under /proc/spl/kstat/zfs.
749 Default value: \fB0\fR.
755 \fBzfs_deadman_enabled\fR (int)
758 Enable deadman timer. See description below.
760 Use \fB1\fR for yes (default) and \fB0\fR to disable.
766 \fBzfs_deadman_synctime_ms\fR (ulong)
769 Expiration time in milliseconds. This value has two meanings. First it is
770 used to determine when the spa_deadman() logic should fire. By default the
771 spa_deadman() will fire if spa_sync() has not completed in 1000 seconds.
772 Secondly, the value determines if an I/O is considered "hung". Any I/O that
773 has not completed in zfs_deadman_synctime_ms is considered "hung" resulting
774 in a zevent being logged.
776 Default value: \fB1,000,000\fR.
782 \fBzfs_dedup_prefetch\fR (int)
785 Enable prefetching dedup-ed blks
787 Use \fB1\fR for yes and \fB0\fR to disable (default).
793 \fBzfs_delay_min_dirty_percent\fR (int)
796 Start to delay each transaction once there is this amount of dirty data,
797 expressed as a percentage of \fBzfs_dirty_data_max\fR.
798 This value should be >= zfs_vdev_async_write_active_max_dirty_percent.
799 See the section "ZFS TRANSACTION DELAY".
801 Default value: \fB60\fR.
807 \fBzfs_delay_scale\fR (int)
810 This controls how quickly the transaction delay approaches infinity.
811 Larger values cause longer delays for a given amount of dirty data.
813 For the smoothest delay, this value should be about 1 billion divided
814 by the maximum number of operations per second. This will smoothly
815 handle between 10x and 1/10th this number.
817 See the section "ZFS TRANSACTION DELAY".
819 Note: \fBzfs_delay_scale\fR * \fBzfs_dirty_data_max\fR must be < 2^64.
821 Default value: \fB500,000\fR.
827 \fBzfs_delete_blocks\fR (ulong)
830 This is the used to define a large file for the purposes of delete. Files
831 containing more than \fBzfs_delete_blocks\fR will be deleted asynchronously
832 while smaller files are deleted synchronously. Decreasing this value will
833 reduce the time spent in an unlink(2) system call at the expense of a longer
834 delay before the freed space is available.
836 Default value: \fB20,480\fR.
842 \fBzfs_dirty_data_max\fR (int)
845 Determines the dirty space limit in bytes. Once this limit is exceeded, new
846 writes are halted until space frees up. This parameter takes precedence
847 over \fBzfs_dirty_data_max_percent\fR.
848 See the section "ZFS TRANSACTION DELAY".
850 Default value: 10 percent of all memory, capped at \fBzfs_dirty_data_max_max\fR.
856 \fBzfs_dirty_data_max_max\fR (int)
859 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed in bytes.
860 This limit is only enforced at module load time, and will be ignored if
861 \fBzfs_dirty_data_max\fR is later changed. This parameter takes
862 precedence over \fBzfs_dirty_data_max_max_percent\fR. See the section
863 "ZFS TRANSACTION DELAY".
865 Default value: 25% of physical RAM.
871 \fBzfs_dirty_data_max_max_percent\fR (int)
874 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed as a
875 percentage of physical RAM. This limit is only enforced at module load
876 time, and will be ignored if \fBzfs_dirty_data_max\fR is later changed.
877 The parameter \fBzfs_dirty_data_max_max\fR takes precedence over this
878 one. See the section "ZFS TRANSACTION DELAY".
880 Default value: \fN25\fR.
886 \fBzfs_dirty_data_max_percent\fR (int)
889 Determines the dirty space limit, expressed as a percentage of all
890 memory. Once this limit is exceeded, new writes are halted until space frees
891 up. The parameter \fBzfs_dirty_data_max\fR takes precedence over this
892 one. See the section "ZFS TRANSACTION DELAY".
894 Default value: 10%, subject to \fBzfs_dirty_data_max_max\fR.
900 \fBzfs_dirty_data_sync\fR (int)
903 Start syncing out a transaction group if there is at least this much dirty data.
905 Default value: \fB67,108,864\fR.
911 \fBzfs_fletcher_4_impl\fR (string)
914 Select a fletcher 4 implementation.
916 Supported selectors are: \fBfastest\fR, \fBscalar\fR, \fBsse2\fR, \fBssse3\fR,
917 \fBavx2\fR, and \fBavx512f\fR.
918 All of the selectors except \fBfastest\fR and \fBscalar\fR require instruction
919 set extensions to be available and will only appear if ZFS detects that they are
920 present at runtime. If multiple implementations of fletcher 4 are available,
921 the \fBfastest\fR will be chosen using a micro benchmark. Selecting \fBscalar\fR
922 results in the original, CPU based calculation, being used. Selecting any option
923 other than \fBfastest\fR and \fBscalar\fR results in vector instructions from
924 the respective CPU instruction set being used.
926 Default value: \fBfastest\fR.
932 \fBzfs_free_bpobj_enabled\fR (int)
935 Enable/disable the processing of the free_bpobj object.
937 Default value: \fB1\fR.
943 \fBzfs_free_max_blocks\fR (ulong)
946 Maximum number of blocks freed in a single txg.
948 Default value: \fB100,000\fR.
954 \fBzfs_vdev_async_read_max_active\fR (int)
957 Maximum asynchronous read I/Os active to each device.
958 See the section "ZFS I/O SCHEDULER".
960 Default value: \fB3\fR.
966 \fBzfs_vdev_async_read_min_active\fR (int)
969 Minimum asynchronous read I/Os active to each device.
970 See the section "ZFS I/O SCHEDULER".
972 Default value: \fB1\fR.
978 \fBzfs_vdev_async_write_active_max_dirty_percent\fR (int)
981 When the pool has more than
982 \fBzfs_vdev_async_write_active_max_dirty_percent\fR dirty data, use
983 \fBzfs_vdev_async_write_max_active\fR to limit active async writes. If
984 the dirty data is between min and max, the active I/O limit is linearly
985 interpolated. See the section "ZFS I/O SCHEDULER".
987 Default value: \fB60\fR.
993 \fBzfs_vdev_async_write_active_min_dirty_percent\fR (int)
996 When the pool has less than
997 \fBzfs_vdev_async_write_active_min_dirty_percent\fR dirty data, use
998 \fBzfs_vdev_async_write_min_active\fR to limit active async writes. If
999 the dirty data is between min and max, the active I/O limit is linearly
1000 interpolated. See the section "ZFS I/O SCHEDULER".
1002 Default value: \fB30\fR.
1008 \fBzfs_vdev_async_write_max_active\fR (int)
1011 Maximum asynchronous write I/Os active to each device.
1012 See the section "ZFS I/O SCHEDULER".
1014 Default value: \fB10\fR.
1020 \fBzfs_vdev_async_write_min_active\fR (int)
1023 Minimum asynchronous write I/Os active to each device.
1024 See the section "ZFS I/O SCHEDULER".
1026 Default value: \fB1\fR.
1032 \fBzfs_vdev_max_active\fR (int)
1035 The maximum number of I/Os active to each device. Ideally, this will be >=
1036 the sum of each queue's max_active. It must be at least the sum of each
1037 queue's min_active. See the section "ZFS I/O SCHEDULER".
1039 Default value: \fB1,000\fR.
1045 \fBzfs_vdev_scrub_max_active\fR (int)
1048 Maximum scrub I/Os active to each device.
1049 See the section "ZFS I/O SCHEDULER".
1051 Default value: \fB2\fR.
1057 \fBzfs_vdev_scrub_min_active\fR (int)
1060 Minimum scrub I/Os active to each device.
1061 See the section "ZFS I/O SCHEDULER".
1063 Default value: \fB1\fR.
1069 \fBzfs_vdev_sync_read_max_active\fR (int)
1072 Maximum synchronous read I/Os active to each device.
1073 See the section "ZFS I/O SCHEDULER".
1075 Default value: \fB10\fR.
1081 \fBzfs_vdev_sync_read_min_active\fR (int)
1084 Minimum synchronous read I/Os active to each device.
1085 See the section "ZFS I/O SCHEDULER".
1087 Default value: \fB10\fR.
1093 \fBzfs_vdev_sync_write_max_active\fR (int)
1096 Maximum synchronous write I/Os active to each device.
1097 See the section "ZFS I/O SCHEDULER".
1099 Default value: \fB10\fR.
1105 \fBzfs_vdev_sync_write_min_active\fR (int)
1108 Minimum synchronous write I/Os active to each device.
1109 See the section "ZFS I/O SCHEDULER".
1111 Default value: \fB10\fR.
1117 \fBzfs_disable_dup_eviction\fR (int)
1120 Disable duplicate buffer eviction
1122 Use \fB1\fR for yes and \fB0\fR for no (default).
1128 \fBzfs_expire_snapshot\fR (int)
1131 Seconds to expire .zfs/snapshot
1133 Default value: \fB300\fR.
1139 \fBzfs_admin_snapshot\fR (int)
1142 Allow the creation, removal, or renaming of entries in the .zfs/snapshot
1143 directory to cause the creation, destruction, or renaming of snapshots.
1144 When enabled this functionality works both locally and over NFS exports
1145 which have the 'no_root_squash' option set. This functionality is disabled
1148 Use \fB1\fR for yes and \fB0\fR for no (default).
1154 \fBzfs_flags\fR (int)
1157 Set additional debugging flags. The following flags may be bitwise-or'd
1169 Enable dprintf entries in the debug log.
1171 2 ZFS_DEBUG_DBUF_VERIFY *
1172 Enable extra dbuf verifications.
1174 4 ZFS_DEBUG_DNODE_VERIFY *
1175 Enable extra dnode verifications.
1177 8 ZFS_DEBUG_SNAPNAMES
1178 Enable snapshot name verification.
1181 Check for illegally modified ARC buffers.
1184 Enable spa_dbgmsg entries in the debug log.
1186 64 ZFS_DEBUG_ZIO_FREE
1187 Enable verification of block frees.
1189 128 ZFS_DEBUG_HISTOGRAM_VERIFY
1190 Enable extra spacemap histogram verifications.
1193 * Requires debug build.
1195 Default value: \fB0\fR.
1201 \fBzfs_free_leak_on_eio\fR (int)
1204 If destroy encounters an EIO while reading metadata (e.g. indirect
1205 blocks), space referenced by the missing metadata can not be freed.
1206 Normally this causes the background destroy to become "stalled", as
1207 it is unable to make forward progress. While in this stalled state,
1208 all remaining space to free from the error-encountering filesystem is
1209 "temporarily leaked". Set this flag to cause it to ignore the EIO,
1210 permanently leak the space from indirect blocks that can not be read,
1211 and continue to free everything else that it can.
1213 The default, "stalling" behavior is useful if the storage partially
1214 fails (i.e. some but not all i/os fail), and then later recovers. In
1215 this case, we will be able to continue pool operations while it is
1216 partially failed, and when it recovers, we can continue to free the
1217 space, with no leaks. However, note that this case is actually
1220 Typically pools either (a) fail completely (but perhaps temporarily,
1221 e.g. a top-level vdev going offline), or (b) have localized,
1222 permanent errors (e.g. disk returns the wrong data due to bit flip or
1223 firmware bug). In case (a), this setting does not matter because the
1224 pool will be suspended and the sync thread will not be able to make
1225 forward progress regardless. In case (b), because the error is
1226 permanent, the best we can do is leak the minimum amount of space,
1227 which is what setting this flag will do. Therefore, it is reasonable
1228 for this flag to normally be set, but we chose the more conservative
1229 approach of not setting it, so that there is no possibility of
1230 leaking space in the "partial temporary" failure case.
1232 Default value: \fB0\fR.
1238 \fBzfs_free_min_time_ms\fR (int)
1241 During a \fRzfs destroy\fB operation using \fRfeature@async_destroy\fB a minimum
1242 of this much time will be spent working on freeing blocks per txg.
1244 Default value: \fB1,000\fR.
1250 \fBzfs_immediate_write_sz\fR (long)
1253 Largest data block to write to zil. Larger blocks will be treated as if the
1254 dataset being written to had the property setting \fRlogbias=throughput\fB.
1256 Default value: \fB32,768\fR.
1262 \fBzfs_max_recordsize\fR (int)
1265 We currently support block sizes from 512 bytes to 16MB. The benefits of
1266 larger blocks, and thus larger IO, need to be weighed against the cost of
1267 COWing a giant block to modify one byte. Additionally, very large blocks
1268 can have an impact on i/o latency, and also potentially on the memory
1269 allocator. Therefore, we do not allow the recordsize to be set larger than
1270 zfs_max_recordsize (default 1MB). Larger blocks can be created by changing
1271 this tunable, and pools with larger blocks can always be imported and used,
1272 regardless of this setting.
1274 Default value: \fB1,048,576\fR.
1280 \fBzfs_mdcomp_disable\fR (int)
1283 Disable meta data compression
1285 Use \fB1\fR for yes and \fB0\fR for no (default).
1291 \fBzfs_metaslab_fragmentation_threshold\fR (int)
1294 Allow metaslabs to keep their active state as long as their fragmentation
1295 percentage is less than or equal to this value. An active metaslab that
1296 exceeds this threshold will no longer keep its active status allowing
1297 better metaslabs to be selected.
1299 Default value: \fB70\fR.
1305 \fBzfs_mg_fragmentation_threshold\fR (int)
1308 Metaslab groups are considered eligible for allocations if their
1309 fragmentation metric (measured as a percentage) is less than or equal to
1310 this value. If a metaslab group exceeds this threshold then it will be
1311 skipped unless all metaslab groups within the metaslab class have also
1312 crossed this threshold.
1314 Default value: \fB85\fR.
1320 \fBzfs_mg_noalloc_threshold\fR (int)
1323 Defines a threshold at which metaslab groups should be eligible for
1324 allocations. The value is expressed as a percentage of free space
1325 beyond which a metaslab group is always eligible for allocations.
1326 If a metaslab group's free space is less than or equal to the
1327 threshold, the allocator will avoid allocating to that group
1328 unless all groups in the pool have reached the threshold. Once all
1329 groups have reached the threshold, all groups are allowed to accept
1330 allocations. The default value of 0 disables the feature and causes
1331 all metaslab groups to be eligible for allocations.
1333 This parameter allows to deal with pools having heavily imbalanced
1334 vdevs such as would be the case when a new vdev has been added.
1335 Setting the threshold to a non-zero percentage will stop allocations
1336 from being made to vdevs that aren't filled to the specified percentage
1337 and allow lesser filled vdevs to acquire more allocations than they
1338 otherwise would under the old \fBzfs_mg_alloc_failures\fR facility.
1340 Default value: \fB0\fR.
1346 \fBzfs_no_scrub_io\fR (int)
1349 Set for no scrub I/O. This results in scrubs not actually scrubbing data and
1350 simply doing a metadata crawl of the pool instead.
1352 Use \fB1\fR for yes and \fB0\fR for no (default).
1358 \fBzfs_no_scrub_prefetch\fR (int)
1361 Set to disable block prefetching for scrubs.
1363 Use \fB1\fR for yes and \fB0\fR for no (default).
1369 \fBzfs_nocacheflush\fR (int)
1372 Disable cache flush operations on disks when writing. Beware, this may cause
1373 corruption if disks re-order writes.
1375 Use \fB1\fR for yes and \fB0\fR for no (default).
1381 \fBzfs_nopwrite_enabled\fR (int)
1386 Use \fB1\fR for yes (default) and \fB0\fR to disable.
1392 \fBzfs_pd_bytes_max\fR (int)
1395 The number of bytes which should be prefetched during a pool traversal
1396 (eg: \fRzfs send\fB or other data crawling operations)
1398 Default value: \fB52,428,800\fR.
1404 \fBzfs_prefetch_disable\fR (int)
1407 This tunable disables predictive prefetch. Note that it leaves "prescient"
1408 prefetch (e.g. prefetch for zfs send) intact. Unlike predictive prefetch,
1409 prescient prefetch never issues i/os that end up not being needed, so it
1410 can't hurt performance.
1412 Use \fB1\fR for yes and \fB0\fR for no (default).
1418 \fBzfs_read_chunk_size\fR (long)
1421 Bytes to read per chunk
1423 Default value: \fB1,048,576\fR.
1429 \fBzfs_read_history\fR (int)
1432 Historic statistics for the last N reads will be available in
1433 \fR/proc/spl/kstat/zfs/POOLNAME/reads\fB
1435 Default value: \fB0\fR (no data is kept).
1441 \fBzfs_read_history_hits\fR (int)
1444 Include cache hits in read history
1446 Use \fB1\fR for yes and \fB0\fR for no (default).
1452 \fBzfs_recover\fR (int)
1455 Set to attempt to recover from fatal errors. This should only be used as a
1456 last resort, as it typically results in leaked space, or worse.
1458 Use \fB1\fR for yes and \fB0\fR for no (default).
1464 \fBzfs_resilver_delay\fR (int)
1467 Number of ticks to delay prior to issuing a resilver I/O operation when
1468 a non-resilver or non-scrub I/O operation has occurred within the past
1469 \fBzfs_scan_idle\fR ticks.
1471 Default value: \fB2\fR.
1477 \fBzfs_resilver_min_time_ms\fR (int)
1480 Resilvers are processed by the sync thread. While resilvering it will spend
1481 at least this much time working on a resilver between txg flushes.
1483 Default value: \fB3,000\fR.
1489 \fBzfs_scan_idle\fR (int)
1492 Idle window in clock ticks. During a scrub or a resilver, if
1493 a non-scrub or non-resilver I/O operation has occurred during this
1494 window, the next scrub or resilver operation is delayed by, respectively
1495 \fBzfs_scrub_delay\fR or \fBzfs_resilver_delay\fR ticks.
1497 Default value: \fB50\fR.
1503 \fBzfs_scan_min_time_ms\fR (int)
1506 Scrubs are processed by the sync thread. While scrubbing it will spend
1507 at least this much time working on a scrub between txg flushes.
1509 Default value: \fB1,000\fR.
1515 \fBzfs_scrub_delay\fR (int)
1518 Number of ticks to delay prior to issuing a scrub I/O operation when
1519 a non-scrub or non-resilver I/O operation has occurred within the past
1520 \fBzfs_scan_idle\fR ticks.
1522 Default value: \fB4\fR.
1528 \fBzfs_send_corrupt_data\fR (int)
1531 Allow sending of corrupt data (ignore read/checksum errors when sending data)
1533 Use \fB1\fR for yes and \fB0\fR for no (default).
1539 \fBzfs_sync_pass_deferred_free\fR (int)
1542 Flushing of data to disk is done in passes. Defer frees starting in this pass
1544 Default value: \fB2\fR.
1550 \fBzfs_sync_pass_dont_compress\fR (int)
1553 Don't compress starting in this pass
1555 Default value: \fB5\fR.
1561 \fBzfs_sync_pass_rewrite\fR (int)
1564 Rewrite new block pointers starting in this pass
1566 Default value: \fB2\fR.
1572 \fBzfs_top_maxinflight\fR (int)
1575 Max concurrent I/Os per top-level vdev (mirrors or raidz arrays) allowed during
1576 scrub or resilver operations.
1578 Default value: \fB32\fR.
1584 \fBzfs_txg_history\fR (int)
1587 Historic statistics for the last N txgs will be available in
1588 \fR/proc/spl/kstat/zfs/POOLNAME/txgs\fB
1590 Default value: \fB0\fR.
1596 \fBzfs_txg_timeout\fR (int)
1599 Flush dirty data to disk at least every N seconds (maximum txg duration)
1601 Default value: \fB5\fR.
1607 \fBzfs_vdev_aggregation_limit\fR (int)
1610 Max vdev I/O aggregation size
1612 Default value: \fB131,072\fR.
1618 \fBzfs_vdev_cache_bshift\fR (int)
1621 Shift size to inflate reads too
1623 Default value: \fB16\fR (effectively 65536).
1629 \fBzfs_vdev_cache_max\fR (int)
1632 Inflate reads small than this value to meet the \fBzfs_vdev_cache_bshift\fR
1635 Default value: \fB16384\fR.
1641 \fBzfs_vdev_cache_size\fR (int)
1644 Total size of the per-disk cache in bytes.
1646 Currently this feature is disabled as it has been found to not be helpful
1647 for performance and in some cases harmful.
1649 Default value: \fB0\fR.
1655 \fBzfs_vdev_mirror_rotating_inc\fR (int)
1658 A number by which the balancing algorithm increments the load calculation for
1659 the purpose of selecting the least busy mirror member when an I/O immediately
1660 follows its predecessor on rotational vdevs for the purpose of making decisions
1663 Default value: \fB0\fR.
1669 \fBzfs_vdev_mirror_rotating_seek_inc\fR (int)
1672 A number by which the balancing algorithm increments the load calculation for
1673 the purpose of selecting the least busy mirror member when an I/O lacks
1674 locality as defined by the zfs_vdev_mirror_rotating_seek_offset. I/Os within
1675 this that are not immediately following the previous I/O are incremented by
1678 Default value: \fB5\fR.
1684 \fBzfs_vdev_mirror_rotating_seek_offset\fR (int)
1687 The maximum distance for the last queued I/O in which the balancing algorithm
1688 considers an I/O to have locality.
1689 See the section "ZFS I/O SCHEDULER".
1691 Default value: \fB1048576\fR.
1697 \fBzfs_vdev_mirror_non_rotating_inc\fR (int)
1700 A number by which the balancing algorithm increments the load calculation for
1701 the purpose of selecting the least busy mirror member on non-rotational vdevs
1702 when I/Os do not immediately follow one another.
1704 Default value: \fB0\fR.
1710 \fBzfs_vdev_mirror_non_rotating_seek_inc\fR (int)
1713 A number by which the balancing algorithm increments the load calculation for
1714 the purpose of selecting the least busy mirror member when an I/O lacks
1715 locality as defined by the zfs_vdev_mirror_rotating_seek_offset. I/Os within
1716 this that are not immediately following the previous I/O are incremented by
1719 Default value: \fB1\fR.
1725 \fBzfs_vdev_read_gap_limit\fR (int)
1728 Aggregate read I/O operations if the gap on-disk between them is within this
1731 Default value: \fB32,768\fR.
1737 \fBzfs_vdev_scheduler\fR (charp)
1740 Set the Linux I/O scheduler on whole disk vdevs to this scheduler
1742 Default value: \fBnoop\fR.
1748 \fBzfs_vdev_write_gap_limit\fR (int)
1751 Aggregate write I/O over gap
1753 Default value: \fB4,096\fR.
1759 \fBzfs_vdev_raidz_impl\fR (string)
1762 Parameter for selecting raidz parity implementation to use.
1764 Options marked (always) below may be selected on module load as they are
1765 supported on all systems.
1766 The remaining options may only be set after the module is loaded, as they
1767 are available only if the implementations are compiled in and supported
1768 on the running system.
1770 Once the module is loaded, the content of
1771 /sys/module/zfs/parameters/zfs_vdev_raidz_impl will show available options
1772 with the currently selected one enclosed in [].
1773 Possible options are:
1774 fastest - (always) implementation selected using built-in benchmark
1775 original - (always) original raidz implementation
1776 scalar - (always) scalar raidz implementation
1777 sse2 - implementation using SSE2 instruction set (64bit x86 only)
1778 ssse3 - implementation using SSSE3 instruction set (64bit x86 only)
1779 avx2 - implementation using AVX2 instruction set (64bit x86 only)
1781 Default value: \fBfastest\fR.
1787 \fBzfs_zevent_cols\fR (int)
1790 When zevents are logged to the console use this as the word wrap width.
1792 Default value: \fB80\fR.
1798 \fBzfs_zevent_console\fR (int)
1801 Log events to the console
1803 Use \fB1\fR for yes and \fB0\fR for no (default).
1809 \fBzfs_zevent_len_max\fR (int)
1812 Max event queue length. A value of 0 will result in a calculated value which
1813 increases with the number of CPUs in the system (minimum 64 events). Events
1814 in the queue can be viewed with the \fBzpool events\fR command.
1816 Default value: \fB0\fR.
1822 \fBzil_replay_disable\fR (int)
1825 Disable intent logging replay. Can be disabled for recovery from corrupted
1828 Use \fB1\fR for yes and \fB0\fR for no (default).
1834 \fBzil_slog_limit\fR (ulong)
1837 Max commit bytes to separate log device
1839 Default value: \fB1,048,576\fR.
1845 \fBzio_delay_max\fR (int)
1848 A zevent will be logged if a ZIO operation takes more than N milliseconds to
1849 complete. Note that this is only a logging facility, not a timeout on
1852 Default value: \fB30,000\fR.
1858 \fBzio_requeue_io_start_cut_in_line\fR (int)
1861 Prioritize requeued I/O
1863 Default value: \fB0\fR.
1869 \fBzio_taskq_batch_pct\fR (uint)
1872 Percentage of online CPUs (or CPU cores, etc) which will run a worker thread
1873 for IO. These workers are responsible for IO work such as compression and
1874 checksum calculations. Fractional number of CPUs will be rounded down.
1876 The default value of 75 was chosen to avoid using all CPUs which can result in
1877 latency issues and inconsistent application performance, especially when high
1878 compression is enabled.
1880 Default value: \fB75\fR.
1886 \fBzvol_inhibit_dev\fR (uint)
1889 Do not create zvol device nodes. This may slightly improve startup time on
1890 systems with a very large number of zvols.
1892 Use \fB1\fR for yes and \fB0\fR for no (default).
1898 \fBzvol_major\fR (uint)
1901 Major number for zvol block devices
1903 Default value: \fB230\fR.
1909 \fBzvol_max_discard_blocks\fR (ulong)
1912 Discard (aka TRIM) operations done on zvols will be done in batches of this
1913 many blocks, where block size is determined by the \fBvolblocksize\fR property
1916 Default value: \fB16,384\fR.
1922 \fBzvol_prefetch_bytes\fR (uint)
1925 When adding a zvol to the system prefetch \fBzvol_prefetch_bytes\fR
1926 from the start and end of the volume. Prefetching these regions
1927 of the volume is desirable because they are likely to be accessed
1928 immediately by \fBblkid(8)\fR or by the kernel scanning for a partition
1931 Default value: \fB131,072\fR.
1934 .SH ZFS I/O SCHEDULER
1935 ZFS issues I/O operations to leaf vdevs to satisfy and complete I/Os.
1936 The I/O scheduler determines when and in what order those operations are
1937 issued. The I/O scheduler divides operations into five I/O classes
1938 prioritized in the following order: sync read, sync write, async read,
1939 async write, and scrub/resilver. Each queue defines the minimum and
1940 maximum number of concurrent operations that may be issued to the
1941 device. In addition, the device has an aggregate maximum,
1942 \fBzfs_vdev_max_active\fR. Note that the sum of the per-queue minimums
1943 must not exceed the aggregate maximum. If the sum of the per-queue
1944 maximums exceeds the aggregate maximum, then the number of active I/Os
1945 may reach \fBzfs_vdev_max_active\fR, in which case no further I/Os will
1946 be issued regardless of whether all per-queue minimums have been met.
1948 For many physical devices, throughput increases with the number of
1949 concurrent operations, but latency typically suffers. Further, physical
1950 devices typically have a limit at which more concurrent operations have no
1951 effect on throughput or can actually cause it to decrease.
1953 The scheduler selects the next operation to issue by first looking for an
1954 I/O class whose minimum has not been satisfied. Once all are satisfied and
1955 the aggregate maximum has not been hit, the scheduler looks for classes
1956 whose maximum has not been satisfied. Iteration through the I/O classes is
1957 done in the order specified above. No further operations are issued if the
1958 aggregate maximum number of concurrent operations has been hit or if there
1959 are no operations queued for an I/O class that has not hit its maximum.
1960 Every time an I/O is queued or an operation completes, the I/O scheduler
1961 looks for new operations to issue.
1963 In general, smaller max_active's will lead to lower latency of synchronous
1964 operations. Larger max_active's may lead to higher overall throughput,
1965 depending on underlying storage.
1967 The ratio of the queues' max_actives determines the balance of performance
1968 between reads, writes, and scrubs. E.g., increasing
1969 \fBzfs_vdev_scrub_max_active\fR will cause the scrub or resilver to complete
1970 more quickly, but reads and writes to have higher latency and lower throughput.
1972 All I/O classes have a fixed maximum number of outstanding operations
1973 except for the async write class. Asynchronous writes represent the data
1974 that is committed to stable storage during the syncing stage for
1975 transaction groups. Transaction groups enter the syncing state
1976 periodically so the number of queued async writes will quickly burst up
1977 and then bleed down to zero. Rather than servicing them as quickly as
1978 possible, the I/O scheduler changes the maximum number of active async
1979 write I/Os according to the amount of dirty data in the pool. Since
1980 both throughput and latency typically increase with the number of
1981 concurrent operations issued to physical devices, reducing the
1982 burstiness in the number of concurrent operations also stabilizes the
1983 response time of operations from other -- and in particular synchronous
1984 -- queues. In broad strokes, the I/O scheduler will issue more
1985 concurrent operations from the async write queue as there's more dirty
1990 The number of concurrent operations issued for the async write I/O class
1991 follows a piece-wise linear function defined by a few adjustable points.
1994 | o---------| <-- zfs_vdev_async_write_max_active
2001 |-------o | | <-- zfs_vdev_async_write_min_active
2002 0|_______^______|_________|
2003 0% | | 100% of zfs_dirty_data_max
2005 | `-- zfs_vdev_async_write_active_max_dirty_percent
2006 `--------- zfs_vdev_async_write_active_min_dirty_percent
2009 Until the amount of dirty data exceeds a minimum percentage of the dirty
2010 data allowed in the pool, the I/O scheduler will limit the number of
2011 concurrent operations to the minimum. As that threshold is crossed, the
2012 number of concurrent operations issued increases linearly to the maximum at
2013 the specified maximum percentage of the dirty data allowed in the pool.
2015 Ideally, the amount of dirty data on a busy pool will stay in the sloped
2016 part of the function between \fBzfs_vdev_async_write_active_min_dirty_percent\fR
2017 and \fBzfs_vdev_async_write_active_max_dirty_percent\fR. If it exceeds the
2018 maximum percentage, this indicates that the rate of incoming data is
2019 greater than the rate that the backend storage can handle. In this case, we
2020 must further throttle incoming writes, as described in the next section.
2022 .SH ZFS TRANSACTION DELAY
2023 We delay transactions when we've determined that the backend storage
2024 isn't able to accommodate the rate of incoming writes.
2026 If there is already a transaction waiting, we delay relative to when
2027 that transaction will finish waiting. This way the calculated delay time
2028 is independent of the number of threads concurrently executing
2031 If we are the only waiter, wait relative to when the transaction
2032 started, rather than the current time. This credits the transaction for
2033 "time already served", e.g. reading indirect blocks.
2035 The minimum time for a transaction to take is calculated as:
2037 min_time = zfs_delay_scale * (dirty - min) / (max - dirty)
2038 min_time is then capped at 100 milliseconds.
2041 The delay has two degrees of freedom that can be adjusted via tunables. The
2042 percentage of dirty data at which we start to delay is defined by
2043 \fBzfs_delay_min_dirty_percent\fR. This should typically be at or above
2044 \fBzfs_vdev_async_write_active_max_dirty_percent\fR so that we only start to
2045 delay after writing at full speed has failed to keep up with the incoming write
2046 rate. The scale of the curve is defined by \fBzfs_delay_scale\fR. Roughly speaking,
2047 this variable determines the amount of delay at the midpoint of the curve.
2051 10ms +-------------------------------------------------------------*+
2067 2ms + (midpoint) * +
2070 | zfs_delay_scale ----------> ******** |
2071 0 +-------------------------------------*********----------------+
2072 0% <- zfs_dirty_data_max -> 100%
2075 Note that since the delay is added to the outstanding time remaining on the
2076 most recent transaction, the delay is effectively the inverse of IOPS.
2077 Here the midpoint of 500us translates to 2000 IOPS. The shape of the curve
2078 was chosen such that small changes in the amount of accumulated dirty data
2079 in the first 3/4 of the curve yield relatively small differences in the
2082 The effects can be easier to understand when the amount of delay is
2083 represented on a log scale:
2087 100ms +-------------------------------------------------------------++
2096 + zfs_delay_scale ----------> ***** +
2107 +--------------------------------------------------------------+
2108 0% <- zfs_dirty_data_max -> 100%
2111 Note here that only as the amount of dirty data approaches its limit does
2112 the delay start to increase rapidly. The goal of a properly tuned system
2113 should be to keep the amount of dirty data out of that range by first
2114 ensuring that the appropriate limits are set for the I/O scheduler to reach
2115 optimal throughput on the backend storage, and then by changing the value
2116 of \fBzfs_delay_scale\fR to increase the steepness of the curve.